[wrangler] fix: don't crash wrangler dev when source-mapping a truncated error chunk by matingathani · Pull Request #14316 · cloudflare/workers-sdk

matingathani · 2026-06-16T05:02:31Z

When a worker throws errors in rapid succession, the stderr chunks received by wrangler dev can arrive mid-frame — so a call site ends up with a negative column number. The @jridgewell/trace-mapping library throws Error: \column` must be greater than or equal to 0` in that case, which was bubbling up uncaught and crashing the wrangler process.

The fix wraps the prepareStack(...) call in getSourceMappedString in a try/catch. If source mapping throws for any reason, we fall back to returning the original (un-source-mapped) string. The worker process keeps running and the user still sees the error message, just without source map translation.

I reproduced this with the snippet from the issue:

export default {
  fetch: () => {
    let obj = {};
    Error.captureStackTrace(obj);
    console.error(obj.stack.slice(0, -3)); // truncated stack
    return new Response('ok');
  },
};

Hitting the worker endpoint a few times caused wrangler dev to crash before this fix. After the fix, it logs the raw stack and keeps running.

Tests
- Additional testing not necessary because: the fix is a defensive try/catch around an existing path. The existing sourcemap tests cover the normal code path. The error path (truncated column) is hard to unit-test deterministically since it depends on OS-level chunk boundaries in stderr delivery.
Public documentation
- Documentation not necessary because: this is a crash fix with no API or behaviour change.

workers-devprod · 2026-06-16T05:03:03Z

Codeowners approval required for this PR:

@cloudflare/wrangler

Show detailed file reviewers

.changeset/fix-miniflare-chrome-corrupt-binary-launch.md: [@cloudflare/wrangler]
.changeset/fix-sourcemap-crash-invalid-column.md: [@cloudflare/wrangler]
packages/deploy-helpers/src/deploy/helpers/sourcemap.ts: [@cloudflare/wrangler]
packages/deploy-helpers/tests/sourcemap.test.ts: [@cloudflare/wrangler]
packages/miniflare/src/plugins/browser-rendering/index.ts: [@cloudflare/wrangler]

- Move for-loop inside try block so sourceMappedStackTrace can be const - Capture caught error and log at debug level instead of silently dropping it

changeset-bot · 2026-06-16T05:08:43Z

🦋 Changeset detected

Latest commit: 52ecf74

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 6 packages

Name	Type
miniflare	Patch
wrangler	Patch
@cloudflare/deploy-helpers	Patch
@cloudflare/pages-shared	Patch
@cloudflare/vite-plugin	Patch
@cloudflare/vitest-pool-workers	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

Wraps the prepareStack call in getSourceMappedString in a try-catch so that if the source map library throws (e.g. because a truncated stderr chunk produced a call site with an invalid column number), the original un-source-mapped string is returned instead of crashing wrangler dev.

pkg-pr-new · 2026-06-16T05:36:09Z

create-cloudflare

npm i https://pkg.pr.new/create-cloudflare@14316

@cloudflare/deploy-helpers

npm i https://pkg.pr.new/@cloudflare/deploy-helpers@14316

@cloudflare/kv-asset-handler

npm i https://pkg.pr.new/@cloudflare/kv-asset-handler@14316

miniflare

npm i https://pkg.pr.new/miniflare@14316

@cloudflare/pages-shared

npm i https://pkg.pr.new/@cloudflare/pages-shared@14316

@cloudflare/unenv-preset

npm i https://pkg.pr.new/@cloudflare/unenv-preset@14316

@cloudflare/vite-plugin

npm i https://pkg.pr.new/@cloudflare/vite-plugin@14316

@cloudflare/vitest-pool-workers

npm i https://pkg.pr.new/@cloudflare/vitest-pool-workers@14316

@cloudflare/workers-auth

npm i https://pkg.pr.new/@cloudflare/workers-auth@14316

@cloudflare/workers-editor-shared

npm i https://pkg.pr.new/@cloudflare/workers-editor-shared@14316

@cloudflare/workers-utils

npm i https://pkg.pr.new/@cloudflare/workers-utils@14316

wrangler

npm i https://pkg.pr.new/wrangler@14316

commit: 52ecf74

matingathani · 2026-06-16T05:46:15Z

The 'Tests (Windows, packages-and-tools)' failure is a pre-existing flake in miniflare's browser tests unrelated to this PR. The Chrome DLL gets locked between parallel test processes on the Windows runner. The same failure appears on main without our changes.

petebacondarwin

I think it would be helpful to have an automated regression test if possible.
Please can you see if you can create one?

…and fix null check - Narrow try-catch to wrap only `prepareStack()` and early-return on failure, so the for-loop replacements are never partially applied then silently dropped - Fix `getFileName() === undefined` → `=== null` (`CallSite.getFileName()` returns `string | null`, never `undefined`, so the old guard always evaluated false) - Add regression test verifying `getSourceMappedString` returns the original value when source mapping fails instead of throwing

matingathani · 2026-06-17T05:01:30Z

Addressed all the review feedback in the latest commit:

Narrowed try-catch to only wrap prepareStack() — instead of wrapping both prepareStack() and the for-loop, the try-catch now only wraps the prepareStack(placeholderError, callSites) call and early-returns the original value on failure. This means the for-loop replacements are never partially applied then silently discarded — either the mapping fully succeeds or the original string is returned unchanged.

Fixed getFileName() === null — CallSite.getFileName() returns string | null, never undefined, so the old === undefined check always evaluated false (filed by Devin). Changed to === null.

Added regression test in packages/deploy-helpers/tests/sourcemap.test.ts — verifies that getSourceMappedString returns the original value without throwing when source mapping fails, covering the truncated-chunk crash scenario.

When Chrome fails to start with "Failed to launch the browser process" (e.g. Windows CI runners serving a DLL that the OS rejects with 0xC1 "%1 is not a valid Win32 application"), delete the stale build directory and retry install + launch once. Complements the existing installWithCorruptedCacheRecovery which handles the "executable missing from cache" case.

devin-ai-integration

Devin Review found 1 new potential issue.

devin-ai-integration · 2026-06-17T05:51:09Z

+"miniflare": patch
+---
+
+fix: recover from corrupt Chrome binary when launching browser for local Browser Rendering


🟡 Changeset title uses prohibited conventional commit prefix

The changeset title on line 5 starts with fix:, which violates the REVIEW.md rule: "Do NOT prefix changeset titles with a 'type' (e.g. fix:, feat:, chore:). The changeset title should be a plain description without conventional commit prefixes." It should read something like "Recover from corrupt Chrome binary when launching browser for local Browser Rendering".

Suggested change

fix: recover from corrupt Chrome binary when launching browser for local Browser Rendering

Recover from corrupt Chrome binary when launching browser for local Browser Rendering

Was this helpful? React with 👍 or 👎 to provide feedback.

…able let Factors the try-catch into a dedicated `tryPrepareStack` helper that returns `null` on failure instead of throwing, letting `getSourceMappedString` use `const` for the result and check `=== null` before entering the for-loop. Addresses Copilot review feedback (prefer const over let-bridging a try block) and aligns with petebacondarwin's suggestion to make prepareStack non-throwing.

When the cached Chrome binary exists but is corrupt, the retry path calls install() again, which fires the downloadProgressCallback and starts the spinner. Without a matching s.stop(), the spinner runs indefinitely through the subsequent launch() and waitForBrowserReady() calls, leaving the terminal in a broken state. Reset startedDownloading before the retry so the callback re-arms cleanly, then stop the spinner after install() returns (success path) or throws (failure path), mirroring the pattern on lines 163-175. Addresses Devin review comment on PR cloudflare#14316.

…n-invalid-column

…umn' into fix/sourcemap-crash-on-invalid-column

petebacondarwin · 2026-06-17T09:39:09Z

@matingathani - this seems to have picked up the changes from another PR too?

Copilot AI review requested due to automatic review settings June 16, 2026 05:02

matingathani requested a review from workers-devprod as a code owner June 16, 2026 05:02

workers-devprod requested review from a team and jamesopstad and removed request for a team June 16, 2026 05:03

This comment was marked as resolved.

Sign in to view

matingathani force-pushed the fix/sourcemap-crash-on-invalid-column branch from 0ee4112 to e7325f9 Compare June 16, 2026 05:24

workers-devprod added this to workers-sdk Jun 16, 2026

github-project-automation Bot moved this to Untriaged in workers-sdk Jun 16, 2026

This comment was marked as resolved.

Sign in to view

petebacondarwin requested changes Jun 16, 2026

View reviewed changes

Comment thread packages/deploy-helpers/src/deploy/helpers/sourcemap.ts Outdated

github-project-automation Bot moved this from Untriaged to In Review in workers-sdk Jun 16, 2026

petebacondarwin removed the request for review from jamesopstad June 16, 2026 09:28

matingathani requested a review from petebacondarwin June 17, 2026 05:08

matingathani and others added 2 commits June 17, 2026 10:38

Merge branch 'main' into fix/sourcemap-crash-on-invalid-column

a09197c

devin-ai-integration Bot reviewed Jun 17, 2026

View reviewed changes

This comment was marked as resolved.

Sign in to view

matingathani and others added 3 commits June 17, 2026 11:36

Merge branch 'main' into fix/sourcemap-crash-on-invalid-column

d499a0c

Merge remote-tracking branch 'origin/main' into fix/sourcemap-crash-o…

41e52c4

…n-invalid-column

Merge remote-tracking branch 'fork/fix/sourcemap-crash-on-invalid-col…

52ecf74

…umn' into fix/sourcemap-crash-on-invalid-column

petebacondarwin marked this pull request as draft June 17, 2026 16:14

	fix: recover from corrupt Chrome binary when launching browser for local Browser Rendering
	Recover from corrupt Chrome binary when launching browser for local Browser Rendering

Conversation

matingathani commented Jun 16, 2026 • edited by devin-ai-integration Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

workers-devprod commented Jun 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment was marked as resolved.

Uh oh!

This comment was marked as resolved.

Uh oh!

changeset-bot Bot commented Jun 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🦋 Changeset detected

Uh oh!

This comment was marked as resolved.

Uh oh!

This comment was marked as resolved.

Uh oh!

pkg-pr-new Bot commented Jun 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

matingathani commented Jun 16, 2026

Uh oh!

petebacondarwin left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

matingathani commented Jun 17, 2026

Uh oh!

devin-ai-integration Bot left a comment

Choose a reason for hiding this comment

Uh oh!

devin-ai-integration Bot Jun 17, 2026

Choose a reason for hiding this comment

Uh oh!

This comment was marked as resolved.

Uh oh!

petebacondarwin commented Jun 17, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

matingathani commented Jun 16, 2026 •

edited by devin-ai-integration Bot

Loading

workers-devprod commented Jun 16, 2026 •

edited

Loading

changeset-bot Bot commented Jun 16, 2026 •

edited

Loading

pkg-pr-new Bot commented Jun 16, 2026 •

edited

Loading